MIT has come out with a paper on AI and government repression that uses a rigorous model to show what most of us already knew through lived experience, intuition and pattern recognition. The abstract concludes:

[C]apital accumulation and thus greater automation encourages the capitalists to support a coup against democracy and set up a repressive system.

It conveniently doesn’t seem to offer any solutions. But it is validating to see the technical reasoning behind the phenomenon we’re witnessing.

Milan professor, Valerio Capraro, summarizes it like this.

As automation shifts income from labor to capital, workers become poorer, revolt becomes more likely, and elites face a choice: redistribute, slow automation, or repress.

But here is the issue: redistribution becomes more expensive precisely when automation makes labor less necessary.

Repression, by contrast, becomes relatively more attractive.

Therefore, elites may prefer repression to slowing down automation or redistributing its gains. This is the dark political economy of AI.

So far, much of the debate has been about: “AI may take your job.”

But economistar Daron Acemoglu, and colleagues Arda Gitmez, and Mehdi Shadmehr make a deeper point: AI may reduce the political need to care whether you have a job.

The future may not be redistribution. It may be surveillance, control, repression.

Full Paper →https://economics.mit.edu/sites/default/files/2026-06/Automation%20and%20Repression.pdf

Valerio’s Linkedin Post → https://www.linkedin.com/posts/valerio-capraro-675235a9_brutal-paper-automation-may-lead-to-political-share-7473400969611874304-FsDd/

George Zeller adds some additional insights on the topic → https://www.linkedin.com/posts/georgzoeller_kind-of-spelling-out-the-obvious-because-share-7473581744387039232-9yS8/